基于用户兴趣差异改进矩阵填充的个性化推荐算法

A PERSONALIZED RECOMMENDATION ALGORITHM BASED ON IMPROVED MATRIX FILLING OF USER INTEREST DIFFERENCE

针对商品推荐领域中传统协同过滤算法受评分数据稀疏性影响导致推荐质量不高的问题,提出一种基于用户兴趣差异的评分矩阵填充方法,优化协同过滤算法的推荐效果。利用电影的关键词和用户选择电影的倾向度改进填充评分矩阵的Slope One算法,对填充后的矩阵预测评分并根据用户的兴趣偏好对结果进行修正。实验结果表明,该算法可以有效地解决传统协同过滤方法在数据稀疏性增大时的问题,提高推荐结果的准确度。

Aiming at the problem of low recommendation quality caused by the sparsity of score data in traditional collaborative filtering algorithm in the field of commodity recommendation,a matrix filling method based on users interests is proposed to optimize the recommendation effects of collaborative filtering algorithm.The Slope One algorithm of filling score matrix was improved by using keywords of movies and propensity of users in choosing movies;the filled matrix was used to predict scores,and the results were modified according to users interests and preferences.Experiments show that the proposed algorithm can effectively solve the problem of the traditional collaborative filtering method when the data sparsity increases,and improve the accuracy of the recommended results.